prof. Ing. Vanda Benešová, CSc.

prof. Ing. Vanda Benešová, CSc.

Ing. Daniel Vašata, Ph.D.

Department of Applied Mathematics

This master's thesis focuses on advanced methods for processing digital histopathological images to improve melanoma cancer prediction using the PUMA dataset. The work focuses on the analysis of current solutions for panoptic segmentation in histopathological data. It proposes a segmentation pipeline in the form of the selected state-of-the-art model TransUnet that is further modified, followed by an Autoencoder for stitching the image patches. This pipeline overcomes the baseline by a 0.06 average DICE score. The thesis includes a detailed description of data preprocessing, hyperparameter optimization, and the implementation of selected models.

Thesis on DSpace

prof. Ing. Vanda Benešová, CSc.

Ing. Magda Friedjungová, Ph.D.

Department of Applied Mathematics

To prevent kidney transplant failure, the condition of the organ must be monitored, and a biopsy is performed when rejection is suspected. Inflammatory cells play an essential role in transplant rejection and their presence and extent need to be evaluated during the assessment of the biopsy, a process that is time-consuming and has limited reproducibility. This thesis focuses on automating this assessment and proposes a deep learning approach for the detection of mononuclear inflammatory cells in kidney transplant biopsies, including their classification into lymphocytes and monocytes. An annotated dataset provided by the MONKEY Challenge, consisting of 81 histological images, was utilized to train the convolutional object detection network YOLO11 using bounding boxes generated from provided dot annotations. The model was evaluated in the MONKEY Challenge and achieved competitive performance among other submissions. The results demonstrate the feasibility of automating the detection process and the proposed model could assist pathologists during kidney transplant biopsy assessment.

Thesis on DSpace

prof. Ing. Vanda Benešová, CSc.

prof. RNDr. Pavel Surynek, Ph.D.

Department of Applied Mathematics

This thesis deals with lesion segmentation of hypoxic-ischemic encephalopathy in neonatal MRI images using deep neural networks. The work explores and implements various approaches such as super resolution and data synthesis to achieve more accurate segmentation on the BONBID-HIE dataset. As part of this work, we implemented a functional pipeline for creating super resolution 3D MRI images, a pipeline for creating synthetic lesions that were further inpainted into the dataset images, and a segmentation pipeline. The results were discussed and compared.

Thesis on DSpace